plastic-labs /
honcho
Memory server that builds lasting profiles of users and agents from conversations
- 7.7k stars
- 309 stars since 3 Oct
- Python
- Updated today
- Some experience
- Memory, RAG & data
Most chatbots forget people between sessions, or replay old text chunks that may not matter. Honcho adds a memory layer. You save messages and events, and Honcho works through them in the background to draw conclusions about each participant.
The data model is simple. A workspace holds peers (users, agents, groups, projects or ideas). Peers take part in sessions, and messages live on sessions. Honcho keeps a representation for each peer. You can ask it questions in plain language, run hybrid search (keyword plus vector), or request prompt-ready context formatted for OpenAI or Anthropic.
It is a FastAPI server. You can use the managed service, run it locally with its CLI, or self-host it. It also ships memory plugins for coding agents such as Claude Code, Codex, Cursor and OpenCode. The README claims strong benchmark results, which you should check on its evals page. Note the AGPL-3.0 license if you self-host.
What it does
Background reasoning over messages
After you add messages to a session, Honcho processes them asynchronously and updates each peer's representation. New messages may take a moment to show up in chat answers.
Peer-based model
Users, agents, groups, projects and ideas are all peers. Honcho can also model what one peer knows about another when configured.
Several ways to read memory
Ask a peer a question with chat, run hybrid BM25 and vector search, fetch a fast static representation, or get session context sized to a token budget.
Coding agent plugins
First-party plugins cover Claude Code, Codex, Cursor, OpenCode, OpenClaw and DeepSeek Harness. Any MCP client can connect too. They share one config file, so agents can share one memory.
Managed, local or self-hosted
Use api.honcho.dev, start a local stack with the honcho CLI, or run the FastAPI server from source with Docker Compose.
What you could build
- Add per-user memory to a tutoring or support chatbot so it remembers learning styles and past problems.
- Give Claude Code, Codex or Cursor persistent memory across coding sessions.
- Share one memory between several agents by pointing them at the same workspace.
- Search past conversations and uploaded documents with hybrid keyword and vector search.
- Inject token-limited session context into an OpenAI or Anthropic call.
Get started
Get an API key at app.honcho.dev, or install the CLI and run honcho start --setup for a local server. Install the SDK, create a peer and session, add messages, then query with chat or context.
pip install honcho-ainpm install @honcho-ai/sdkCommands are copied from the project's README. Read it on GitHub for the full setup.
More in Memory, RAG & data
firecrawl
Web API that turns sites and search results into markdown or JSON for LLMs
markitdown
Python tool that turns PDFs, Office files, audio and more into Markdown for LLMs
graphify
Maps a codebase and its docs into a knowledge graph your AI assistant can query
claude-mem
Saves what your coding agent does and feeds the useful parts back into later sessions
vectorize-io/hindsight
Memory server that lets AI agents store, search and learn from past interactions
OpenViking
A context database that stores agent memory, docs and skills as browsable files
open-notebook
Self-hosted research notebook that chats with your documents using any AI provider
WeKnora
Self-hosted platform that turns documents into RAG search, an agent and a wiki
supermemory
Memory layer that lets AI assistants and apps remember users across conversations
gbrain
A shared memory for AI agents that stores sourced facts and answers with citations